intermediate results. ix. detection of the galactic haze ... · planck collaboration: detection of...

arX

iv:1

208.

5483

v1 [

astr

o-ph

.GA

] 27

Aug

201

2Astronomy & Astrophysicsmanuscript no. planck˙haze c© ESO 2012August 29, 2012

Planck intermediate results. IX. Detection of the Galactic haze withPlanck

Planck Collaboration: P. A. R. Ade83, N. Aghanim57, M. Arnaud72, M. Ashdown68,5, F. Atrio-Barandela16, J. Aumont57, C. Baccigalupi82,A. Balbi33, A. J. Banday88,7, R. B. Barreiro64, J. G. Bartlett1,66, E. Battaner89, K. Benabed58,86, A. Benoıt55, J.-P. Bernard7, M. Bersanelli30,47,A. Bonaldi67, J. R. Bond6, J. Borrill11,84, F. R. Bouchet58,86, C. Burigana46,32, P. Cabella34, J.-F. Cardoso73,1,58, A. Catalano74,71, L. Cayon27,

R.-R. Chary54, L.-Y Chiang60, P. R. Christensen79,35, D. L. Clements53, L. P. L. Colombo20,66, A. Coulais71, B. P. Crill66,80, F. Cuttaia46,L. Danese82, O. D’Arcangelo65, R. J. Davis67, P. de Bernardis29, G. de Gasperis33, A. de Rosa46, G. de Zotti42,82, J. Delabrouille1, C. Dickinson67,

J. M. Diego64, G. Dobler69, H. Dole57,56, S. Donzelli47, O. Dore66,8, U. Dorl77, M. Douspis57, X. Dupac37, G. Efstathiou61, T. A. Enßlin77,H. K. Eriksen62, F. Finelli46, O. Forni88,7, M. Frailis44, E. Franceschi46, S. Galeotta44, K. Ganga1, M. Giard88,7, G. Giardino38,

J. Gonzalez-Nuevo64,82, K. M. Gorski66,91⋆, S. Gratton68,61, A. Gregorio31,44, A. Gruppuso46, F. K. Hansen62, D. Harrison61,68, G. Helou8,S. Henrot-Versille70, C. Hernandez-Monteagudo10,77, S. R. Hildebrandt8, E. Hivon58,86, M. Hobson5, W. A. Holmes66, A. Hornstrup14,W. Hovest77, K. M. Huffenberger90, T. R. Jaffe88,7, T. Jagemann37, W. C. Jones22, M. Juvela21, E. Keihanen21, J. Knoche77, L. Knox24,M. Kunz15,57, H. Kurki-Suonio21,40, G. Lagache57, A. Lahteenmaki2,40, J.-M. Lamarre71, A. Lasenby5,68, C. R. Lawrence66, S. Leach82,

R. Leonardi37, P. B. Lilje62,9, M. Linden-Vørnle14, M. Lopez-Caniego64, P. M. Lubin25, J. F. Macıas-Perez74, B. Maffei67, D. Maino30,47,N. Mandolesi46,4, M. Maris44, P. G. Martin6, E. Martınez-Gonzalez64, S. Masi29, M. Massardi45, S. Matarrese28, F. Matthai77, P. Mazzotta33,P. R. Meinhold25, A. Melchiorri29,48, L. Mendes37, A. Mennella30,47, S. Mitra52,66, M.-A. Miville-Deschenes57,6, A. Moneti58, L. Montier88,7,

G. Morgante46, D. Munshi83, J. A. Murphy78, P. Naselsky79,35, P. Natoli32,3,46, H. U. Nørgaard-Nielsen14, F. Noviello67, S. Osborne85, F. Pajot57,R. Paladini54, D. Paoletti46, B. Partridge39, T. J. Pearson8,54, O. Perdereau70, F. Perrotta82, F. Piacentini29, M. Piat1, E. Pierpaoli20, D. Pietrobon66,S. Plaszczynski70, E. Pointecouteau88,7, G. Polenta3,43, N. Ponthieu57,50, L. Popa59, T. Poutanen40,21,2, G. W. Pratt72, S. Prunet58,86, J.-L. Puget57,J. P. Rachen18,77, R. Rebolo63,12,36, M. Reinecke77, C. Renault74, S. Ricciardi46, T. Riller77, G. Rocha66,8, C. Rosset1, J. A. Rubino-Martın63,36,

B. Rusholme54, M. Sandri46, G. Savini81, B. M. Schaefer87, D. Scott19, G. F. Smoot23,76,1, F. Stivoli49, R. Sudiwala83, A.-S. Suur-Uski21,40,J.-F. Sygnet58, J. A. Tauber38, L. Terenzi46, L. Toffolatti17,64, M. Tomasi47, M. Tristram70, M. Turler51, G. Umana41, L. Valenziano46, B. Van

Tent75, P. Vielva64, F. Villa46, N. Vittorio33, L. A. Wade66, B. D. Wandelt58,86,26, M. White23, D. Yvon13, A. Zacchei44, and A. Zonca25

(Affiliations can be found after the references)

Preprint online version: August 29, 2012

ABSTRACT

Using precise full-sky observations fromPlanck, and applying several methods of component separation, we identify and characterize the emissionfrom the Galactic “haze” at microwave wavelengths. The hazeis a distinct component of diffuse Galactic emission, roughly centered on the Galacticcentre, and extends to|b| ∼ 35◦ in Galactic latitude and|l| ∼ 15◦ in longitude. By combining thePlanck data with observations from theWilkinsonMicrowave Anisotropy Probe we are able to determine the spectrum of this emission to highaccuracy, unhindered by the large systematic biasespresent in previous analyses. The derived spectrum is consistent with power-law emission with a spectral index of−2.55± 0.05, thus excludingfree-free emission as the source and instead favouring hard-spectrum synchrotron radiation from an electron population with a spectrum (numberdensity per energy)dN/dE ∝ E−2.1. At Galactic latitudes|b| < 30◦, the microwave haze morphology is consistent with that of theFermi gamma-ray“haze” or “bubbles,” indicating that we have a multi-wavelength view of a distinct component of our Galaxy. Given both the very hard spectrumand the extended nature of the emission, it is highly unlikely that the haze electrons result from supernova shocks in theGalactic disk. Instead, anew mechanism for cosmic-ray acceleration in the centre of our Galaxy is implied.

Key words. Galaxy: nucleus – ISM: structure – ISM: bubbles – radio continuum: ISM

1. Introduction

The initial data release from theWilkinson MicrowaveAnisotropy Probe (WMAP) revolutionised our understanding ofboth cosmology (Spergel et al. 2003) and the physical processesat work in the interstellar medium (ISM) of our own Galaxy(Bennett et al. 2003). Some of the processes observed wereexpected, such as the thermal emission from dust grains, free-free emission (or thermal bremsstrahlung) from electron/ionscattering, and synchrotron emission due to shock-acceleratedelectrons interacting with the Galactic magnetic field. Others,such as the anomalous microwave emission now identified as

⋆ Corresponding author: K. M. Gorski, e-mail:[email protected]

spinning dust emission from rapidly rotating tiny dust grains(Draine & Lazarian 1998a,b; de Oliveira-Costa et al. 2002;Finkbeiner et al. 2004; Hinshaw et al. 2007; Boughn & Pober2007; Dobler & Finkbeiner 2008b; Dobler et al. 2009), weremore surprising. But perhaps most mysterious was a “haze” ofemission discovered byFinkbeiner (2004a) that was centredon the Galactic centre (GC), appeared roughly sphericallysymmetric in profile, fell off roughly as the inverse distancefrom the GC, and was of unknown origin. This haze wasoriginally characterised as free-free emission byFinkbeiner(2004a) due to its apparently very hard spectrum, although itwas not appreciated at the time how significant the systematicuncertainty in the measured spectrum was.

1

http://arxiv.org/abs/1208.5483v1

[email protected]

Planck Collaboration: Detection of the Galactic haze withPlanck

An analysis of the 3-year WMAP data byDobler & Finkbeiner (2008a, hereafter DF08) identified asource of systematic uncertainty in the determination of thehaze spectrum that remains the key to determining the originof the emission. This uncertainty is due to residual foregroundscontaminating the cosmic microwave background (CMB) radi-ation estimate used in the analysis, and arises as a consequenceof chance morphological correlations between the CMB and thehaze itself. Nevertheless, the spectrum was found to be bothsig-nificantly softer than free-free emission, and also significantlyharder than the synchrotron emission observed elsewhere intheGalaxy as traced by the low-frequency synchrotron measure-ments ofHaslam et al.(1982) (see alsoReich & Reich 1988;Davies et al. 1996; Kogut et al. 2007; Strong et al. 2011; Kogut2012). Finally, it was noted that this systematic uncertaintycould be almost completely eliminated with data from thePlanck1 mission, which would produce estimates of the CMBsignal that were significantly less contaminated by Galacticforegrounds.

The synchrotron nature of the microwave haze was substan-tially supported by the discovery of a gamma-ray counterpart tothis emission byDobler et al.(2010) using data from theFermiGamma-Ray Space Telescope. These observations were consis-tent with an inverse Compton (IC) signal generated by electronswith the same spectrum and amplitude as would yield the mi-crowave haze atWMAP wavelengths. Further work bySu et al.(2010) showed that theFermi haze appeared to have sharpedges and it was renamed the “Fermi bubbles.” Subsequently,there has been significant theoretical interest in determiningthe origin of the very hard spectrum of progenitor electrons.Suggestions include enhanced supernova rates (Biermann et al.2010), a Galactic wind (Crocker & Aharonian 2011), a jet gen-erated by accretion onto the central black hole (Guo & Mathews2011; Guo et al. 2011), and co-annihilation of dark matter (DM)particles in the Galactic halo (Finkbeiner 2004b; Hooper et al.2007; Lin et al. 2010; Dobler et al. 2011). However, while eachof these scenarios can reproduce some of the properties of thehaze/bubbles well, none can completely match all of the ob-served characteristics (Dobler 2012).

Moreover, despite the significant observational evidence,there have been suggestions in the literature that the mi-crowave haze is either an artefact of the analysis proce-dure (Mertsch & Sarkar 2010) or not synchrotron emission(Gold et al. 2011). The former conclusion was initially sup-ported by alternative analyses of theWMAP data that foundno evidence of the haze (Eriksen et al. 2006; Dickinson et al.2009). However, more recentlyPietrobon et al.(2012) showedthat these analyses, while extremely effective at cleaning theCMB of foregrounds and identifying likely contaminants ofa known morphology (e.g., a low-level residual cosmologicaldipole), typically cannot separate the haze emission from alow-frequency combination of free-free, spinning dust, and softersynchrotron radiation. The argument ofGold et al.(2011) thatthe microwave haze is not synchrotron emission was based onthe lack of detection of a polarised component. This criticismwas addressed byDobler (2012) who showed that, even if theemission is not depolarised by turbulence in the magnetic field,

1 Planck (http://www.esa.int/Planck) is a project of theEuropean Space Agency (ESA) with instruments provided by two sci-entific consortia funded by ESA member states (in particularthe leadcountries France and Italy), with contributions from NASA (USA) andtelescope reflectors provided by a collaboration between ESA and a sci-entific consortium led and funded by Denmark.

such a polarised signal is not likely to be seen withWMAP giventhe noise in the data.

With the Planck data, we now have the ability not only toprovide evidence for the existence of the microwave haze withan independent experiment, but also to eliminate the uncertaintyin the spectrum of the emission which has hindered both obser-vational and theoretical studies for nearly a decade. In Sect.2 wedescribe thePlanck data as well as some external templates weuse in our analysis. In Sect.3 we describe the two most effectivecomponent separation techniques for studying the haze emissionin temperature. In Sect.4 we discuss our results on the morphol-ogy and spectrum of the haze, before summarising in Sect.5.

2. Planck data and templates

Planck (Tauber et al. 2010; Planck Collaboration I 2011) is thethird generation space mission to measure the anisotropy ofthecosmic microwave background (CMB). It observes the sky innine frequency bands covering 30–857GHz with high sensitiv-ity and angular resolution from 31′ to 5′. The Low FrequencyInstrument (LFI;Mandolesi et al. 2010; Bersanelli et al. 2010;Mennella et al. 2011) covers the 30, 44, and 70 GHz bands withamplifiers cooled to 20 K. The High Frequency Instrument (HFI;Lamarre et al. 2010; Planck HFI Core Team 2011a) covers the100, 143, 217, 353, 545, and 857 GHz bands with bolome-ters cooled to 0.1 K. Polarisation is measured in all but thehighest two bands (Leahy et al. 2010; Rosset et al. 2010). Acombination of radiative cooling and three mechanical cool-ers produces the temperatures needed for the detectors and op-tics (Planck Collaboration II 2011). Two data processing centres(DPCs) check and calibrate the data and make maps of the sky(Planck HFI Core Team 2011b; Zacchei et al. 2011). Planck’ssensitivity, angular resolution, and frequency coverage make it apowerful instrument for galactic and extragalactic astrophysicsas well as cosmology. Early astrophysics results are given inPlanck Collaboration VIII–XXVI 2011, based on data taken be-tween 13 August 2009 and 7 June 2010. Intermediate astro-physics results are now being presented in a series of papersbased on data taken between 13 August 2009 and 27 November2010.

We take both theWMAP andPlanck bandpasses into accountwhen defining our central frequencies. However, throughoutwerefer to the bands by the conventional labels of 23, 33, 41, 61,and 94 GHz forWMAP and 30, 44, 70, 100, 143, 217, 353, 545,and 857 GHz forPlanck; the central frequencies are 22.8, 33.2,41.0, 61.4, and 94.0 GHz, and 28.5, 44.1, 70.3, 100.0, 143.0,217.0, 353.0, 545.0, and 857.0 GHz respectively. In each case,the central frequency represents the convolution of the bandpassresponse with a CMB spectrum and so corresponds to the ef-fective frequency for emission with that spectrum. For emissionwith different spectra, the effective frequency is slightly shifted,but the effects are at the few percent level and do not significantlyaffect our conclusions.

Our analysis also requires the use of external templates tomorphologically trace emission mechanisms within thePlanckdata. All the data are available in the HEALPix2 scheme(Gorski et al. 2005). In each case, we use maps smoothed to 1◦

angular resolution.

Thermal and spinning dust For a template of the combinedthermal and spinning dust emission, we use the 100µm all-

2 see http://healpix.jpl.nasa.gov

2

http://www.esa.int/Planck


sky map fromSchlegel et al.(1998) evaluated at the appro-priate Planck and WMAP frequencies using Model 8 fromFinkbeiner et al.(1999, FDS99). This is a sufficiently good esti-mate of the thermal emission for our purposes, although it isim-portant to note that the morphological correlation betweenther-mal and spinning dust is not well known.

Free-free The free-free template adopted in our analysis is theHα map assembled byFinkbeiner(2003)3 from three surveys:the Wisconsin Hα Mapper (Haffner et al. 2003), the SouthernHα Sky Survey Atlas (Gaustad et al. 2001), and the VirginiaTech Spectral-Line Survey (Dennison et al. 1998). The map iscorrected for line-of-sight dust absorption assuming uniformmixing between gas and dust, although we mask some regionsbased on the predicted total dust extinction where the correctionto the Hα emission is deemed unreliable.

Soft Synchrotron Since synchrotron intensity rises with de-creasing frequency, the 408 MHz full-sky radio continuum map(Haslam et al. 1982) provides a reasonable tracer of the soft syn-chrotron emission. While there is a very small contributionfromfree-free emission to the observed intensity, particularly in theGalactic plane, the bulk of the emission traces synchrotronra-diation from supernova shock-accelerated electrons that havehad sufficient time to diffuse from their source. In addition, aspointed out byDobler(2012), the propagation length for cosmic-ray electrons in the disk is energy-dependent and thereforethe408 MHz map (which is dominated by synchrotron emissionfrom lower energy electrons compared to the situation at 20–100 GHz) will be more spatially extended than the synchrotronatPlanck frequencies (seeMertsch & Sarkar 2010). This can re-sult in a disk-like residual when using the 408 MHz map as atracer of synchrotron at higher frequencies that could be con-fused with the haze emission. We use an elliptical Gaussian disktemplate (σl = 20◦ andσb = 5◦) for this residual, though inpractice this results in only a very small correction to our results,which use a larger mask thanDobler(2012) (see below).

The Haze Although a measurement of the precise morphologyof the microwave haze is to be determined, an estimate of themorphology is necessary to reduce bias in template fits for thefollowing reason: when using templates to separate foregrounds,the amplitudes of the other templates may be biased to com-pensate for the haze emission present in the data unless an ap-propriate haze template is used to approximate the emission.FollowingDobler(2012), we use an elliptical Gaussian templatewith σl = 15◦ andσb = 25◦. Note that a map of theFermigamma-ray haze/bubbles cannot be used to trace the emissionfor two reasons. First, as pointed out byDobler et al.(2011),the morphology of the gamma-ray emission is uncertain at lowlatitudes. Second, the synchrotron morphology depends sensi-tively on the magnetic field while the gamma-ray morphologydepends on the interstellar radiation field. Therefore, while thesame cosmic-ray population is clearly responsible for both, thedetailed morphologies are not identical.4

3 Our specific choice of theFinkbeiner(2003) Hα template does nothave a strong impact on results. We have repeated our analysis using theDickinson et al.(2003) Hα map and find differences at the few percentlevel that are not spatially correlated with haze emission.

4 We have performed our fits using the uniform “bubbles” templategiven in Su et al.(2010) and the morphology of the haze excess (seeSect.4) is not significantly changed.

Mask As noted above, the effect of dust extinction requirescareful treatment of the Hαmap when using it as a tracer of free-free emission. Therefore, we mask out all regions where dustex-tinction at Hα wavelengths is greater than 1 mag. We also maskout all point sources in theWMAP and Planck ERCSC (30–143 GHz) catalogs. Several larger-scale features where ourtem-plates are likely to fail are also masked: the LMC, SMC, M31,Orion–Barnard’s Loop, NGC 5128, andζ Oph. Finally, since theHα to free-free ratio is a function of gas temperature, we maskpixels with Hα intensity greater than 10 rayleigh to minimise thebias due to strong spatial fluctuations in gas temperatures.Thismask covers 32% of the sky and is shown in Fig.1.

3. Component separation methods

In this paper, we apply two methods for separating the Galacticemission components in thePlanck data. The first one, used inthe originalWMAP haze analyses, is a simple regression tech-nique in which the templates described in the previous sectionare fit directly to the data. This “template fitting” method isrelatively simple to implement and its results are easy to inter-pret. Furthermore, the noise characteristics are well understoodand additional components not represented by the templatesarereadily identifiable in residual maps. The second technique, apowerful power-spectrum estimation and component-separationmethod based on Gibbs sampling, uses a Bayesian approach andcombines pixel-by-pixel spectral fits with template amplitudes.One of the significant advantages of this approach is that, ratherthan assuming an estimate for the CMB anisotropy, a CMB mapis generated via joint sampling of the foreground parameters andCℓs of cosmological anisotropies; this should reduce the biasinthe inferred foreground spectra.

3.1. Template fitting

The rationale behind the simple template fitting technique isthat there are only a few physical mechanisms in the interstel-lar medium that generate emission at microwave wavelengths,and these emission mechanisms are morphologically traced bymaps at other frequencies at which they dominate. We fol-low the linear regression formalism ofFinkbeiner (2004a),Dobler & Finkbeiner(2008a), andDobler (2012) and solve therelation

dν = aν · P, (1)

wheredν is a data map at frequencyν, P is a matrix of the tem-plates defined in Sect.2, andaν is the vector of scaling ampli-tudes for this set of templates. The least-squares solutionto thisequation is

aν = (PTN−1ν P)−1(PTN−1

ν dν), (2)

whereNν is the noise covariance matrix at frequencyν. In prac-tice, for our template fits we use the mean noise per band (i.e.,we setNν = 〈Nν〉 for all pixels), which is appropriate in thelimit where the dominant uncertainty is how well the templatestrace the foregrounds, as is the case here. To the extent thatthetemplates morphologically match the actual foregrounds, the so-lutions ai

ν for templatei as a function of frequency represent areasonable estimate of the spectrum over the fitted pixels.

There are two important features of this approach to templatefitting that must be addressed. First, there is an implicit assump-tion that the spectrum of a given template-correlated emission

3


mechanism does not vary across the region of interest, and sec-ond, an estimate for the CMB must be pre-subtracted from thedata. The former can be validated by inspecting a map of theresiduals which can reveal where this assumption fails, andasa consequence of which the sky can easily be subdivided intoregions that can be fitted independently. The latter involves thecomplication that no CMB estimate is completely clean of theforegrounds to be measured, which thereore introduces a bias(with the same spectrum as the CMB) in the inferred foregroundspectra. As shown byDF08, this bias becomes increasingly largewith frequency and renders an exact measurement of the hazespectrum impossible withWMAP alone. This “CMB bias” isthe dominant source of uncertainty in all foreground analyses.However,DF08 also pointed out that, because the haze spec-trum falls with frequency, the high-frequency data fromPlanckcan be used to generate a CMB estimate that is nearly completelyfree from haze emission. Thus, pre-subtraction of this estimateshould result in an essentially unbiased estimate of the hazespectrum. The CMB estimate that we use consists of a “PlanckHFI internal linear combination” (PILC) map, formed from aminimum-variance linear combination of thePlanck HFI 143–545 GHz data after pre-subtraction of the thermal dust modelofFDS99at each frequency.5 Defining pν and tν to be thePlanckmaps andFDS99prediction (respectively) at frequencyν, thePILC in∆TCMB is given by

PILC = 1.39× (p143− t143) − 0.36× (p217− t217)− 0.025× (p353− t353) + 0.0013× (p545− t545).

(3)

The weights are determined by minimising the the variance overunmasked pixels of the PILC while maintaining a unity responseto the CMB spectrum.

Although no constraint is made on the spectral dependenceof the template coefficients in Eq.2, the fit does assume thatthe spectrum is constant across the sky. While this assumptionis actually quite good outside our mask (as we show below), itis known to be insufficient in detail. As such, in addition to full(unmasked) sky fits, we also perform template fits on smallersky regions and combine the results to form a full compositemap. The subdivisions are defined by hand to separate the skyinto regions with particularly large residuals in a full-sky fit andare listed in Table1.

3.2. Gibbs sampling: Commander

An alternative method for minimising the CMB bias is to gener-ate a CMB estimate from the data while simultaneously solvingfor the parameters of a Galactic foreground model. Within theBayesian framework it is possible to set stronger priors on theCMB parameterisation (i.e.,Cℓs), taking advantage not only ofthe frequency spectrum of the CMB (a blackbody), but also ofthe angular power spectrum of the fluctuations. Even for rela-tively simple foreground models, the dimensionality of param-eter space is quite large so uniform sampling on a grid is notfeasible.

5 Pre-subtracting theFDS99prediction for the thermal dust is notmeant to provide a perfect model for the thermal dust, but rather a rea-sonable model. The goal is to minimise variance in the PILC and itis more effective to do so by pre-subtracting the dust model. This al-lows the fit to manage the CO contamination present at variousHFI fre-quency channels more effectively (although there is still some leakagehowever, see Sect.4.1). We have tested a PILC which does not subtractthe thermal dust and the morphology and amplitude of the recoveredhaze signal are similar.

Table 1. Regions used for the multi-region (RG) template fits.

Region Sky Coverage

1 −125◦ ≤ l < −104◦ −30◦ ≤ b < 0◦

2 −104◦ ≤ l < −80◦ −30◦ ≤ b < 0◦

3 −125◦ ≤ l < −104◦ 0◦ ≤ b < 30◦

4 −104◦ ≤ l < −80◦ 0◦ ≤ b < 30◦

5 −37◦ ≤ l < 42◦ 0◦ ≤ b < 90◦

6 −80◦ ≤ l < −25◦ −30◦ ≤ b < 0◦

7 70◦ ≤ l < 180◦ −90◦ ≤ b < 0◦

8 12◦ ≤ l < 70◦ −90◦ ≤ b < 0◦

9 Unmasked pixels outside regions 1–8 andb ≤ 010 Unmasked pixels outside regions 1–8 andb > 0

Jewell et al. (2004) and Wandelt et al. (2004) first dis-cussed the application of Gibbs sampling algorithms (a vari-ant of MCMC sampling) in this context. These algorithmshave been further improved (Eriksen et al. 2004; O’Dwyer et al.2004; Eriksen et al. 2007; Chu et al. 2005; Jewell et al. 2009;Rudjord et al. 2009; Larson et al. 2007) and packaged into theCommander code.

Gibbs sampling is particularly suitable for component sep-aration since it samples from the conditional distributionalongperpendicular directions in parameter space, updating thedis-tribution with each sample. This approach has been advo-cated byEriksen et al.(2007, 2008a) andDickinson et al.(2009)and has been applied recently to theWMAP 7-year data byPietrobon et al.(2012). A detailed description of the algorithmand its validation on simulated data is provided byEriksen et al.(2008b, and references therein).

The outputs of the sampling are a map-based CMB estimateand the parameters of a foreground model, which can either betemplate-based, pixel-based, or a combination of the two. Weperform the analysis at HEALPix resolutionNside = 128. Thechoice of the foreground model is limited by the number of fre-quency channels observed since it sets the number of constraintson the model when fitting spectra for each pixel. We separateour results in the following section into two categories, fits usingPlanck data only and fits usingPlanck data plus ancillary datasets.

For thePlanck-only fits, our model consists of a single powerlaw T ∝ ν βS describing the effective low-frequency emission(with a prior on spectral index,βS = −3.05 ± 0.3), a grey-body for the thermal dust emission that dominates at high fre-quencies (with a temperature and emissivity prior given by theresults ofPlanck Collaboration XIX 2011, where mean valuesof TD ≃ 18 K andǫD = 1.8 were measured), and a CO spec-trum. The CO spectrum is assumed constant across the sky andnormalised to 100 GHz. The relative strength of theJ=2→1(∼ 217 GHz) andJ=3→2 (∼ 353 GHz) transition lines with re-spect to theJ=1→0 transition were computed by taking into ac-count the specifications of the HFI detectors and calibratedbymeans of the available survey (Dame et al. 2001). The relativeratios in the 100, 217, and 353 GHz bands are 1.0, 0.35, and 0.12respectively. We checked the robustness of the result against aplausible variation of the line ratios of∼ 10%. (A more detaileddiscussion of the CO analysis that we performed can be foundin Planck Collaboration XIX 2011). We normalise the thermaldust component at 353 GHz and the low-frequency power lawat 33 GHz. Hence, we solve for two spectral indices togetherwith the corresponding amplitudes as well as a CO amplitude,with the dust temperature fixed at a value of 18 K. The cur-

4


rentCommander implementation allows for the determination ofresidual monopole and dipole contributions, as may result fromthe calibration and map-making procedures. This fit is referredto as CMD1 throughout. It is interesting to note that, given thenoise in the data, this highly over-simplified model is sufficientto describe the total Galactic emission (see Sect.4.1). However,it is well established that the low-frequency emission actu-ally consists of several components. FollowingPietrobon et al.(2012), our procedure for separating these components is to per-form a template fit as specified in Eq.2 on theCommander solu-tion for the low-frequency amplitude (i.e., replacingdν with thelow-frequency amplitude map).Pietrobon et al.(2012) showedthat applying this “post-processing” template regressionproce-dure is effective in extracting the haze from theCommander so-lution.

The addition of theWMAP channels allows us to refinethe foreground model further, separating the multiple contribu-tions in the frequency range 23–70 GHz. Moreover, the inclusionof the 408 MHz data improves the characterisation of the syn-chrotron component and will allow us to investigate the spatialvariations of its spectral index (see Sect.4.2). TheCommanderfit, CMD2, is then based on 14 frequency maps (eightPlanckchannels from 30 to 353 GHz, five fromWMAP, and Haslam408 MHz), and allows a modification of the foreground modelto encompass two low-frequency power-law components – onesoft component with a fixed spectral indexβS = −3.05 to de-scribe the soft synchrotron emission6 and one with a spectralindexβH with prior βH = −2.15± 0.3 to capture both the hardsynchrotron haze and the free-free emission. With this model,the low-frequency part of the spectrum is more easily resolvedinto physically meaningful components.

In addition, we parameterise a joint thermal and spinning-dust model by

Djd(ν) =

(

ν

ν0

)1+ǫ B(ν, T )B(ν0, T )

+ eαe−[(ν−ν1)/b]2/2. (4)

This is the sum of a grey-body spectrum for the thermal dust,and a Gaussian profile to mimic the spinning dust SED. The lat-ter is a purely phenomenological model selected on the basisof its straightforward numerical implementation. However, wehave established its effectiveness in describing well-known spin-ning dust regions in the Gould Belt (Planck Intermediate Paper,in preparation). The thermal dust pivot frequencyν0 is set to545 GHz and the spinning dust peak frequencyν1 to 20 GHz.The remaining parameters (the amplitude of the joint spectrum,the relative amplitude of the spinning dust contribution, and thewidth of the spinning dust bump) are constrained by the Gibbssampling procedure. As before we also adopt a spectrum for theCO emission.

4. Results

In what follows, we perform four different types of haze extrac-tion:

1. A masked full-sky (FS) template fit for each input frequencyband.

6 This value represents the spectral index of the large Loop I fea-ture that is a prominent supernova remnant visible at both 408 MHz andmicrowave frequencies in the northern Galactic hemisphere. We haverepeated our analysis varying this index byδβ = 0.1 and find no signif-icant difference in our results.

2. Template fits over subsections of the sky (RG) that are com-bined to give a full-sky haze map for each input frequencyband.

3. A Commander fit (CMD1) with a simple two-componentforeground model, usingPlanck 30–353GHz data.

4. A comprehensiveCommander fit (CMD2) including thermaland spinning dust models, a soft power-law component, anda hard power-law component, usingPlanck 30–545GHz,WMAP 23–94GHz, and Haslam 408 MHz data sets.

We first discuss our results from the template fitting andGibbs sampling analyses derived from thePlanck data alone,then proceed to include external data sets in the analysis. Adirect comparison of the results between the template fits andCommander haze extraction methods boosts confidence that, notonly are components being appropriately separated, but thespec-trum is relatively free from bias.

4.1. Planck-only results

4.1.1. Template fitting

Figure1 presents the templates and mask used for thePlanckanalysis, together with the CMB-subtracted data and best fittem-plate model at 30 GHz. We also show the full-sky (i.e., unre-stricted inl andb) haze residual, defined as

RHν = dν − aν · P + aH

ν · h, (5)

where h is the haze template defined in Sect.2. The haze isclearly present in thePlanck data set and, as illustrated in Fig.2(left column), scaling each residual byν2.5 yields roughly equalbrightness per frequency band indicating that the spectrumisapproximatelyT H

ν ∝ ν−2.5. A more detailed measurement of

the spectrum will be given in Sect.4.3. It is also interesting tonote that the morphology does not change significantly with fre-quency (although striping in thePlanck HFI maps used to formthe CMB estimate is a significant contaminant at frequenciesabove∼ 40 GHz) indicating that the spectrum of the haze emis-sion is roughly constant with position.

The haze residual is most clearly visible in the southernGC region, but we note that our assumption of uniform spectraacross the sky does leave some residuals around the edge of themask and in a few particularly bright free-free regions. However,while our imperfect templates and assumptions about uniformspectra have done a remarkable job of isolating the haze emis-sion (96% of the total variance is removed in the fit atPlanck30 GHz), we can more effectively isolate the haze by subdivid-ing the sky into smaller regions as described in Sect.3.1. Theresultant full-sky haze residual is shown in Fig.2. With this fit,the residuals near the mask are cleaner and we have done a bet-ter job in fitting the difficult Ophiucus region in the northern GC,though striping again becomes a major contaminant for frequen-cies above∼ 40 GHz.

4.1.2. Commander

Figure3 presents the results of our CMD1Commander fit andthe subsequent post-processing. As noted previously, thisverysimple model provides an adequate description of the data witha meanχ2 of 18.4 (7 d.o.f.) outside the mask, despite the factthat the low-frequency component is really an aggregate of sev-eral different emission mechanisms, as shown byPietrobon et al.(2012). It is visually apparent that the low-frequency amplitudeis highly correlated with thermal dust emission in some regions,

5


Haslam Hα FDS 30 GHz

Haze Template Disk Template Mask

30 GHz Model 30 GHz Planck 30 GHz Haze Residual

-0.1 0.2Tant × (ν/23 GHz)2.5 [mK]

Fig. 1. The templates and full-sky template fitting model (see Sect.4.1). Top left: the Haslam et al.(1982) 408 MHz map.Topmiddle: theFinkbeiner(2003) Hαmap.Top right: theFinkbeiner et al.(1999) dust prediction at thePlanck 30 GHz channel.Middleleft: the elliptical Gaussian haze template.Center: the elliptical Gaussian disk template.Middle right: the mask used in the fit.Bottom left: the best fit template linear combination model atPlanck 30 GHz.Bottom middle: the CMB-subtractedPlanck data at30 GHz.Bottom right: the Planck 30 GHz data minus the 30 GHz model with the haze template component added back into themap.

suggesting a dust origin for some of this emission (e.g., spinningdust). Finally, features that are well known from low-frequencyradio surveys, such as Loop I, are also visible, implying a syn-chrotron origin, with a spectral index closer toβS = −3. Thecoefficients of the post-processing template-based fit describedin Sect.3.2are given in Table2 and show a strong positive cor-relation with each template.

As with the template fitting case, we see from Fig.3 that thepost-processing residuals for the low-frequency CMD1 compo-nent are low except towards the Galactic centre where the haze isclearly present, implying that it is emission with a distinct mor-phology compared to the dust, free-free, and soft synchrotronemission. Furthermore, the morphology is strikingly similar tothe template fitting indicating strong consistency betweentheresults. Since an analogous regression cannot be performedonthe spectral-index map, a more flexible foreground model mustbe implemented to isolate the haze spectrum. However, the ad-ditional model parameters require the use of external data sets.

4.2. Results from Planck plus external data sets

4.2.1. Template fitting

In order to further our understanding of the spectrum and mor-phology of the microwave haze component, we augment thePlanck data with theWMAP 7-year data set (covering thefrequency range 23–94 GHz) and the 408 MHz data. For thetemplate-fitting method, the inclusion of the new data is triv-ial since Eq.2 does not assume anything about the frequencydependence of the spectrum and each map is fit independently.The results for the full sky and for smaller regional fits are shownin Figs.4 and5. The haze residual is present in both theWMAPandPlanck data, and the morphology and spectrum appear con-sistent between data sets. As before, scaling each residualbyν2.5 yields roughly equal brightness per band from 23 GHz to61 GHz. Including theWMAP data also confirms that the mor-phology does not change significantly with frequency, thus im-plying a roughly constant haze spectrum with position.

6


-0.05

0.10T

ant ×

(ν/

23 G

Hz)

2.5 [m

K]

30 GHz Planck haze (FS)

-0.05

0.10

Tant × (ν/23 G

Hz) 2.5 [m

K]

30 GHz Planck haze (RG)

-0.05

0.10

Tan

t × (

ν/23

GH

z)2.

5 [mK

]


-0.05

0.10

Tant × (ν/23 G

Hz) 2.5 [m

K]


Fig. 2. Left column: thePlanck haze (i.e., the same as the bottom right panel of Fig.1), for thePlanck 30 and 44 GHz channels usinga full-sky template fit to the data. A scaling ofν2.5 yields roughly equal brightness residuals indicating thatthe haze spectrum isroughlyTν ∝ ν−2.5, implying that the electron spectrum is a very harddN/dEe ∝ E−2. Note that the haze appears more elongated inlatitude than longitude by a factor of two, which is roughly consistent with theFermi gamma-ray haze/bubbles (Dobler et al. 2010).For frequencies above∼ 40 GHz, striping in the HFI channels (which contaminates ourCMB estimate) begins to dominate over thehaze emission.Right column: the same but for the “regional” fits described in Sect.4.1. The overall morphology of the haze is thesame, but the residuals near the mask and in the Ophiucus complex in the north GC are improved.

Table 2. Regression coefficients of theCommander foreground amplitude maps.

Fit coefficientFit type Data sets

Hα [mK/R] FDS [mK/mK] Haslam [mK/K] Haze [mK/arbitrary]

CMD1 Planck 30–353 GHz 2.8× 10−3 ± 2.0× 10−4 1.9± 4.3× 10−2 1.6× 10−6 ± 4.4× 10−8 6.0× 10−2 ± 3.4× 10−3

CMD2 Planck 30–353 GHz,WMAP, Haslam

3.3× 10−3 ± 3.9× 10−4 1.0± 8.4× 10−2 2.4× 10−9 ± 8.8× 10−8 5.7× 10−2 ± 6.7× 10−3

4.2.2. Commander

Comparing the low frequency, hard spectral indexCommandersolution at 23 GHz obtained with this model to our previous(less flexible) parameterisation, we find that the residualscor-related with the Haslam 408 MHz map are significantly reducedas shown in Fig.3. Table2 lists the fit coefficients in this case,and we now find no significant correlation with the Haslam map.As before, a template regression illustrates that the haze residualis significant and our hard spectrum power law contains bothfree-free and haze emission.7 Furthermore, Fig.6 illustrates thatthe fixedβS = −3.05 power law provides a remarkably goodfit to the 408 MHz data. Indeed, subtracting this soft-spectrumcomponent from the map yields nearly zero residuals outside

7 A close comparison between the CMD1 and CMD2 results suggeststhat the haze amplitude is slightly lower in the latter. However, due tothe flexibility of the CMD2 model (specifically the fact that the modelallows for the unphysical case of non-zero spinning dust in regions ofnegligible thermal dust), it is likely that some of the haze emission isbeing included in the spinning dust component.

the mask, except for bright free-free regions which contaminatethe Haslam et al.(1982) map at the∼ 10% level. It is interest-ing to note that this residual (as well as the negligible Haslam-correlation coefficient in Table2) imply that fits assuming a con-stant spectral index across the sky for this correlated emissionare reasonable. Physically, this means that electrons do diffuse toa steady-state spectrum which is very close todN/dE ∝ E−3 (inagreement with the propagation models ofStrong et al. 2011).

Taken together, Figs.3 and 6 imply that, not only is the408 MHz-correlated soft synchrotron emission consistent with aspectral index of−3.05 across the entire sky (outside our mask)from 408 MHz to 60 GHz, but the haze region consists of both asoft and a hard component. That is, the haze is not a simple vari-ation of spectral index from 408 MHz to∼ 20 GHz. If it were,then our assumption ofβS = −3.05 (i.e., the wrong spectral in-dex for the haze) would yield residuals in the difference map ofFig.6. The map of the harder spectral index would ideally be adirect measurement of the haze spectrum. However, the signal-to-noise ratio is only sufficient to accurately measure the spec-trum in the very bright free-free regions (e.g., the Gum Nebula).

7


-0.1

0.2T

ant x

(ν/

23 G

Hz)

2.5 [m

K]

Low Frequency Amplitude (CMD1)

-0.1

0.2

Tant x (ν/23 G

Hz) 2.5 [m

K]

Hard Spectrum Amplitude (CMD2)

-0.1

0.2

Tan

t x (

ν/23

GH

z)2.

5 [mK

]

Low Frequency Template Model (CMD1)

-0.1

0.2

Tant x (ν/23 G

Hz) 2.5 [m

K]

Hard Spectrum Template Model (CMD2)

-0.05

0.10

Tan

t x (

ν/23

GH

z)2.

5 [mK

]

Low Frequency Residual (CMD1)

-0.05

0.10

Tant x (ν/23 G

Hz) 2.5 [m

K]

Hard Spectrum Residual (CMD2)

Fig. 3. Left column, top: The recovered amplitude of the low-frequency component at23 GHz from our simplestCommanderfit to the Planck data alone, CMD1. As shown inPietrobon et al.(2012), while this model provides an excellent description ofthe data, this low-frequency component is actually a combination of free-free, spinning dust, and synchrotron emission (top).Left column, middle: a four-component template model of this component (see Table 2). Left column, bottom: The haze residual.The residuals are small outside the haze region indicating that the templates are a reasonable morphological representation of thedifferent components contained in theCommander solution. The haze residual is strikingly similar to that found for the template-only approach in Fig.2 (though there does seem to be a residual dipole in theCommander solution).Right column: The same, but forthe CMD2 low-frequency, hard spectrum component. While there is still some leakage of dust-correlated emission in the solution,the softer synchrotron emission (mostly correlated with the 408 MHz template [see Fig.6]) has been separated byCommander. Theresultant map is dominated by free-free and the haze emission and the regressed haze residual (bottom panel) shows morphologyvery similar to both the template fitting and CMD1 results indicating that the haze has been effectively isolated.

In the fainter haze region, the spectral index is dominated bynoise in the maps.

4.3. Spectrum and morphology

While a pixel-by-pixel determination of the haze spectrum is notpossible given the relatively low signal-to-noise ratio per pixelof the haze emission, we can get a reliable estimate of its meanbehaviour from the template fitting residuals in Fig.5. The ma-jority of previous haze studies have estimated the haze spectrumvia the template coefficientsaν for the haze template. However,as noted inDobler(2012), such an estimate is not only affectedby the CMB bias (which we have effectively minimised by us-ing the PILC), but may also be biased by the effect of imperfecttemplate morphologies. The argument is as follows: consider a

perfectly CMB-subtracted map which consists of the true hazeh′ plus another true foreground componentf ′ which we are ap-proximating by templatesh and f respectively. Our template fitapproach can be written as

aHh + aF f = bHh′ + bF f ′, (6)

where we are solving foraH andaF while bH andbF are the trueamplitudes. TheaH solution to this equation is

aH = bH ×Γhh′ − Γ f h′Γh f

1− Γ f hΓh f+ bF ×

Γh f ′ − Γ f f ′Γh f

1− Γ f hΓh f, (7)

where, for example,Γh f ′ ≡ 〈h f ′〉/〈h2〉, and the mean is overunmasked pixels. Thus, ifh = h′ and f = f ′ thenaH = bH andwe recover the correct spectrum. However, ifh , h′ then the

8


-0.05

0.10T

ant ×

(ν/

23 G

Hz)

2.5 [m

K]

23 GHz WMAP haze (FS)

-0.05

0.10

Tant × (ν/23 G

Hz) 2.5 [m

K]


-0.05

0.10

Tan

t × (

ν/23

GH

z)2.

5 [mK

]


-0.05

0.10

Tant × (ν/23 G

Hz) 2.5 [m

K]


-0.05

0.10

Tan

t × (

ν/23

GH

z)2.

5 [mK

]


-0.05

0.10

Tant × (ν/23 G

Hz) 2.5 [m

K]


Fig. 4. The microwave haze at bothWMAP andPlanck wavelengths using a full-sky template fit to the data. The morphology ofthe haze is remarkably consistent from band to band and between data sets implying that the spectrum of the haze does not varysignificantly with position. Furthermore, theν2.5 scaling again yields roughly equal-brightness residuals indicating that the hazespectrum is roughlyTν ∝ ν−2.5 through both thePlanck andWMAP channels. In addition, while striping is minimally important atlow frequencies, above∼40 GHz it becomes comparable to, or brighter than, the haze emission (see text).

spectrum is biased and iff , f ′ it is biased and dependent uponthe true spectrum of the other foreground,bF.

We emphasise that this bias is dependent on the cross-correlation of the true foregrounds with the templates (which isunknown) and that we have assumed a perfectly clean CMB esti-mate (which is not possible to create) and have not discussedtheimpact of striping or other survey artefacts (which Figs.4 and5show are present). Given this, a much more straightforward es-timate of the haze spectrum is to measure it directly fromRH ina region that is relatively devoid of artefacts or other emission.We measure the spectrum in the GC south region|l| < 35◦ and−35◦ < b < 0◦ by performing a linear fit (slope and offset) overunmasked pixels and convert the slope measurement to a powerlaw given the central frequencies of thePlanck andWMAP data(see Fig.7). Specifically, we fit

R23H = Aν × R

νH + Bν (8)

over unmasked pixels in this region forAν andBν, and calculatethe haze spectral index,βH = log(Aν)/ log(ν/23 GHz), for eachν. This spectrum should now be very clean and – given our useof the PILC – reasonably unbiased.

A measurement of the spectrum of the haze emission isshown in Fig.7. It is evident that theWMAP andPlanck bandsare complementarily located in log-frequency space and thetwo experiments together provide significantly more informationthan either one alone.8 In the left panel we plot〈RνH〉−Bν (wherethe mean is over the unmasked pixels in the region given aboveand the errors are their standard deviation). The haze spectrum ismeasured to beTν ∝ ν βH with βH = −2.55± 0.05. This spectrumis a nearly perfect power law from 23 to 41 GHz. Furthermore,if we form the total synchrotron residual,

RS = RH + aS · s, (9)

where s is the Haslam map, and measure its spectrum in thesouth GC, we again recover a nearly perfect power law withβS = −3.1. Our conclusion is that the haze, which is not con-sistent with free-free emission, arises from synchrotron emis-

8 The close log-frequency spacing of theWMAP 94 GHz andPlanck100 GHz channels has the significant advantage that the CO (J=1→0)line falls in thePlanck 100 GHz band while it is outside theWMAP94 GHz band. This provides an excellent estimate for the CO morphol-ogy.

9


-0.05

0.10T

ant ×

(ν/

23 G

Hz)

2.5 [m

K]

23 GHz WMAP haze (RG)

-0.05

0.10

Tant × (ν/23 G

Hz) 2.5 [m

K]


-0.05

0.10

Tan

t × (

ν/23

GH

z)2.

5 [mK

]


-0.05

0.10

Tant × (ν/23 G

Hz) 2.5 [m

K]


-0.05

0.10

Tan

t × (

ν/23

GH

z)2.

5 [mK

]


-0.05

0.10

Tant × (ν/23 G

Hz) 2.5 [m

K]


Fig. 5. The same as Fig.4 but using the regions defined inDF08. Clearly, the residuals near the mask are significantly reduced,although, as with the full-sky fits, striping in the HFI channels (which leaks into the CMB estimate) becomes significant above∼ 40 GHz.

-22

45

0.40

8 G

Hz

Tan

tenn

a [K

]

Soft Spectrum Amplitude (CMD2)

-2.2

4.5

0.408 GH

z Tantenna [K

]

Haslam Minus Soft Spectrum (CMD2)

Fig. 6. Left: The soft synchrotron component at 408 MHz from theCommanderCMD2 analysis. The map is strikingly similar to theHaslam map (see Fig.1) indicating that soft synchrotron emission has a very uniform spectrum from 408 MHz to 60 GHz throughall of the data sets.Right: The difference between the Haslam map and theCommander solution. This is consistent with noise acrossalmost the entire sky with the exception of a few bright free-free clouds that are present in the Haslam data at the∼ 10% level. Thelack of significant haze emission in the difference map (particularly in the south) is a strong indication that the haze region consistsof both a hardand a soft component rather than having a simple spatially variable spectral index.

sion with a spectral index that is harder than elsewhere in theGalaxy byβH − βS = 0.5. Within the haze region, this compo-nent represents∼ 33% of the total synchrotron and 23% of thetotal Galactic emission at 23 GHz (WMAP K-band) while emis-

sions correlated with Haslam, Hα, and FDS contribute 43%, 4%,and 30% respectively.

TheβH = −2.55 spectral index of the haze is strongly indica-tive of synchrotron emission from a population of electronswith

10


20 30 40 50ν [GHz]

0.04

0.06

0.08

0.10

0.20

Tan

tenn

a ×

(ν/2

3 G

Hz)

2 [mK

]

total synch3 × hazeIν ∝ ν-0.55

Iν ∝ ν-1.10

-0.2 0.0 0.2 0.4 0.6T23

antenna [mK]

-0.1

0.0

0.1

0.2

0.3

T30 an

tenn

a [m

K]

total synch, βS = -3.07haze only, βH = -2.58

-0.2 0.0 0.2 0.4 0.6T23

antenna [mK]

-0.1

0.0

0.1

0.2

0.3

T33 an

tenn

a [m

K]

total synch, βS = -3.09haze only, βH = -2.55

Fig. 7. Left: The spectrum measured from the residual in Fig.5 in the region|l| < 25◦, −35◦ < b < −10◦. The haze spectrum isvery nearly a power law with spectral indexβH = −2.55, while the total synchrotron emission in the region has a spectral indexof βS = −3.1 (see Sect.4.3), significantlysofter than the haze emission. This spectrum should be free from biases due to templateuncertainties.Middle andright: Scatter plots (shown in contours) for both the haze (dotted) and total synchrotron (solid) emissionusingWMAP 23–33 GHz andPlanck 30 GHz.

a spectrum that is harder than elsewhere in the Galaxy. The otherpossible origins of the emission in this frequency range (namely,free-free and spinning dust) are strongly disfavored for severalreasons. First, the spinning dust mechanism is very unlikelysince there is no corresponding feature in thermal dust emissionat HFI frequencies. While it is true that environment can havean impact on both the grain size distribution and relative ratio ofspinning to thermal dust emission (thus making the FDS modelsan imperfect tracer of spinning dust, e.g.,Ysard et al. 2011), togenerate a strong spinning dust signal at LFI frequencies whilenot simultaneously producing a thermal signal a highly contrivedgrain population would be required, in which small grains sur-vive but large grains are completely destroyed. Furthermore, theFDS thermal predictions yield very low dust-correlated residuals(see Fig.5) indicating a close correspondence between thermaland spinning-dust morphology. Finally, this spectrum is signif-icantly softer than free-free emission, which has a characteris-tic spectral index≈ −2.15. Since the Hα to free-free ratio istemperature-dependent, the possibility exists that the haze emis-sion represents some mixture of synchrotron and free-free with-out yielding a detectable Hα signal. However, in order to have ameasured spectral index ofβH ≈ −2.5 from 23 to 41 GHz, free-free could only represent 50% of the emission if the synchrotroncomponent had a spectral index≈ −3. Since such a steep spectralindex is ruled out by the lack of a strong haze signal at 408 MHz,the synchrotron emission must have a harder spectrum and thefree-free component (if it exists) must be subdominant.9 Theseconsiderations, coupled with the likely inverse-Compton signalwith Fermi (seeDobler et al. 2010; Su et al. 2010), strongly in-dicate a separate component of synchrotron emission.

4.4. Spatial correspondence with the Fermi haze/bubbles

The gamma-ray emission from theFermi haze/bubbles(Dobler et al. 2010; Su et al. 2010) is consistent with the inverse-Compton emission from a population of electrons with the en-

9 In addition, the lack of a bremsstrahlung signal in X-rays requiresa fine tuning of the gas temperature to be∼ 106 K, a temperature atwhich the gas has a very short cooling time. This also argues against afree-free explanation as described inMcQuinn & Zaldarriaga(2011).

ergy spectrum required to reproduce theβH = 2.55 haze emis-sion measured in this paper. Furthermore, theFermi “haze” hasa very strong spatial coincidence with thePlanck microwaves atlow latitude (below|b| ∼ 35◦) as we show in Fig.8. This suggestsa common physical origin for these two measurements with thegamma-ray contribution extending down tob ≈ −50◦, whilethe microwaves fall off quickly belowb ≈ −35◦. As in Dobler(2012), the interpretation is that the magnetic field within thehaze/bubbles sharply decreases above∼ 5 kpc from the Galacticplane while the cosmic-ray distribution extends to∼ 10 kpcand continues to generate gamma-ray emission (e.g., by inverseCompton scattering CMB photons). In Fig.9 we show a full-sky representation of thePlanck haze emission overlaid with theFermi gamma-ray haze/bubbles fromDobler et al.(2010).

5. Summary

We have identified the presence of a microwave haze in thePlanck LFI data and performed a joint analysis with 7-yearWMAP data. Our findings verify not only that the haze is real,but also that it is consistent in amplitude and spectrum in thesetwo different experiments. Furthermore, we have usedPlanckHFI maps to generate a CMB estimate that is nearly completelyclean of haze emission, implying that we have reduced system-atic biases in the inferred spectrum to a negligible level. We findthat the unbiased haze spectrum is consistent with a power lawof spectral indexβH = −2.55± 0.05, ruling out free-free emis-sion as a possible explanation, and strengthening the possibil-ity of a hard synchrotron component origin. The spectrum ofsofter synchrotron emission found elsewhere in the Galaxy isβS = −3.1, consistent with a cosmic-ray electron population thathas been accelerated in supernova shocks and diffused through-out the Galaxy. This spectrum is significantly softer than thehaze emission, which is not consistent with supernova shockac-celeration after taking into account energy losses from diffusioneffects.

The microwave haze is detected in thePlanck maps withboth simple template regression against the data and a moresophisticated Gibbs sampling analysis. The former provides anexcellent visualisation of the haze at each wavelength on largescales while the latter allows a pixel-by-pixel analysis ofthe

11


45 0 -45-90

0

-35

45 0 -45-90

0

-35

-0.012 0.041Thermo ∆T [mK]

Fig. 8. Left: The southernPlanck 30 GHz haze from Fig.5. Right: The same but with contours of theFermi gamma-ray haze/bubbles(Su et al. 2010) overlaid in white. Aboveb = −35◦ (orange dashed line), the morphological correspondence isvery strong suggestingthat the two signals are generated by the same underlying phenomenon.

complete data set. While the template analysis allows us to de-rive theβH = −2.55 spectrum with high confidence, spectral de-termination with the Gibbs approach is more difficult given thatnoise must be added to the analysis to ensure convergence in thesampling method, and that a significantly more flexible model(in particular, one in which the spectrum of synchrotron is al-lowed to vary with each pixel) is used. However, not only is thespatial correspondence of the haze derived with the two methodsexcellent, but the Gibbs method allows us to show conclusivelythat the microwave haze is a separate component and not merelya variation in the spectral index of the synchrotron emission.

The morphology of the microwave haze is nearly identicalfrom 23 to 44 GHz, implying that the spectrum does not varysignificantly with position. Although detection of the hazein po-larisation withWMAP remains unlikely given the noise level ofthe data (Dobler 2012), future work withPlanck will concentrateon using its enhanced sensitivity to search for this component.

Acknowledgements. The development ofPlanck has been supported by: ESA;CNES and CNRS/INSU-IN2P3-INP (France); ASI, CNR, and INAF (Italy);NASA and DoE (USA); STFC and UKSA (UK); CSIC, MICINN and JA(Spain); Tekes, AoF and CSC (Finland); DLR and MPG (Germany); CSA(Canada); DTU Space (Denmark); SER/SSO (Switzerland); RCN (Norway);SFI (Ireland); FCT/MCTES (Portugal); and DEISA (EU). A description ofthe Planck Collaboration and a list of its members, including the technicalor scientific activities in which they have been involved, can be found athttp://www.rssd.esa.int/Planck. G. Dobler has been supported by theHarvey L. Karp Discovery Award. Some of the results in this paper have beenderived using the HEALPix (Gorski et al. 2005) package.

ReferencesBennett, C. L., Hill, R. S., Hinshaw, G., et al. 2003, ApJS, 148, 97Bersanelli, M., Mandolesi, N., Butler, R. C., et al. 2010, A&A, 520, A4+Biermann, P. L., Becker, J. K., Caceres, G., et al. 2010, ApJ,710, L53Boughn, S. P. & Pober, J. C. 2007, ApJ, 661, 938Chu, M., Eriksen, H. K., Knox, L., et al. 2005, Phys. Rev. D, 71, 103002

Crocker, R. M. & Aharonian, F. 2011, Phys. Rev. Lett., 106, 101102Dame, T. M., Hartmann, D., & Thaddeus, P. 2001, ApJ, 547, 792Davies, R. D., Watson, R. A., & Gutierrez, C. M. 1996, MNRAS, 278, 925de Oliveira-Costa, A., Tegmark, M., Finkbeiner, D. P., et al. 2002, ApJ, 567, 363Dennison, B., Simonetti, J. H., & Topasna, G. A. 1998, PASA, 15, 147Dickinson, C., Davies, R. D., & Davis, R. J. 2003, MNRAS, 341,369Dickinson, C., Eriksen, H. K., Banday, A. J., et al. 2009, ApJ, 705, 1607Dobler, G. 2012, ApJ, 750, 17Dobler, G., Cholis, I., & Weiner, N. 2011, ApJ, 741, 25Dobler, G., Draine, B., & Finkbeiner, D. P. 2009, ApJ, 699, 1374Dobler, G. & Finkbeiner, D. P. 2008a, ApJ, 680, 1222Dobler, G. & Finkbeiner, D. P. 2008b, ApJ, 680, 1235Dobler, G., Finkbeiner, D. P., Cholis, I., Slatyer, T., & Weiner, N. 2010, ApJ,

717, 825Draine, B. T. & Lazarian, A. 1998a, ApJ, 494, L19Draine, B. T. & Lazarian, A. 1998b, ApJ, 508, 157Eriksen, H. K., Dickinson, C., Jewell, J. B., et al. 2008a, ApJ, 672, L87Eriksen, H. K., Dickinson, C., Lawrence, C. R., et al. 2006, ApJ, 641, 665Eriksen, H. K., Huey, G., Saha, R., et al. 2007, ApJ, 656, 641Eriksen, H. K., Jewell, J. B., Dickinson, C., et al. 2008b, ApJ, 676, 10Eriksen, H. K., O’Dwyer, I. J., Jewell, J. B., et al. 2004, ApJS, 155, 227Finkbeiner, D. P. 2003, ApJS, 146, 407Finkbeiner, D. P. 2004a, ApJ, 614, 186Finkbeiner, D. P. 2004b, arXiv:astro-ph/0409027Finkbeiner, D. P., Davis, M., & Schlegel, D. J. 1999, ApJ, 524, 867Finkbeiner, D. P., Langston, G. I., & Minter, A. H. 2004, ApJ,617, 350Gaustad, J. E., McCullough, P. R., Rosing, W., & Van Buren, D.2001, PASP,

113, 1326Gold, B., Odegard, N., Weiland, J. L., et al. 2011, ApJS, 192,15Gorski, K. M., Hivon, E., Banday, A. J., et al. 2005, ApJ, 622, 759Guo, F. & Mathews, W. G. 2011, arXiv:1103.0055Guo, F., Mathews, W. G., Dobler, G., & Oh, S. P. 2011, arXiv:1110.0834Haffner, L. M., Reynolds, R. J., Tufte, S. L., et al. 2003, ApJS, 149, 405Haslam, C. G. T., Salter, C. J., Stoffel, H., & Wilson, W. E. 1982, A&AS, 47, 1Hinshaw, G., Nolta, M. R., Bennett, C. L., et al. 2007, ApJS, 170, 288Hooper, D., Finkbeiner, D. P., & Dobler, G. 2007, Phys. Rev. D, 76, 083012Jewell, J., Levin, S., & Anderson, C. H. 2004, ApJ, 609, 1Jewell, J. B., Eriksen, H. K., Wandelt, B. D., et al. 2009, ApJ, 697, 258Kogut, A. 2012, ApJ, 753, 110Kogut, A., Dunkley, J., Bennett, C. L., et al. 2007, ApJ, 665,355Lamarre, J., Puget, J., Ade, P. A. R., et al. 2010, A&A, 520, A9+

12

http://www.rssd.esa.int/Planck


Fig. 9. Top: The microwave haze atPlanck 30 GHz (red,−12µK < ∆TCMB < 30 µK) and 44 GHz (yellow, 12µK < ∆TCMB < 40µK). Bottom: The same but including theFermi 2-5 GeV haze/bubbles ofDobler et al.(2010) (blue, 1.05 < intensity [keV cm−2

s−1 sr−1] < 1.25; see their Fig. 11). The spatial correspondence between the two is excellent, particularly at low southern Galacticlatitude, suggesting that this is a multi-wavelength view of the same underlying physical mechanism.

Larson, D. L., Eriksen, H. K., Wandelt, B. D., et al. 2007, ApJ, 656, 653Leahy, J. P., Bersanelli, M., D’Arcangelo, O., et al. 2010, A&A, 520, A8+Lin, T., Finkbeiner, D. P., & Dobler, G. 2010, Phys. Rev. D, 82, 023518Mandolesi, N., Bersanelli, M., Butler, R. C., et al. 2010, A&A, 520, A3+McQuinn, M. & Zaldarriaga, M. 2011, MNRAS, 414, 3577Mennella et al. 2011, A&A, 536, A3Mertsch, P. & Sarkar, S. 2010, JCAP, 10, 19O’Dwyer, I. J., Eriksen, H. K., Wandelt, B. D., et al. 2004, ApJ, 617, L99Pietrobon, D., Gorski, K. M., Bartlett, J., et al. 2012, ApJ, 755, 69Planck Collaboration I. 2011, A&A, 536, A1Planck Collaboration II. 2011, A&A, 536, A2Planck Collaboration XIX. 2011, A&A, 536, 19Planck HFI Core Team. 2011a, A&A, 536, A4Planck HFI Core Team. 2011b, A&A, 536, A6

Reich, P. & Reich, W. 1988, A&AS, 74, 7Rosset, C., Tristram, M., Ponthieu, N., et al. 2010, A&A, 520, A13+Rudjord, Ø., Groeneboom, N. E., Eriksen, H. K., et al. 2009, ApJ, 692, 1669Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525Spergel, D. N., Verde, L., Peiris, H. V., et al. 2003, ApJS, 148, 175Strong, A. W., Orlando, E., & Jaffe, T. R. 2011, A&A, 534, A54Su, M., Slatyer, T. R., & Finkbeiner, D. P. 2010, ApJ, 724, 1044Tauber, J. A., Mandolesi, N., Puget, J., et al. 2010, A&A, 520, A1+Wandelt, B. D., Larson, D. L., & Lakshminarayanan, A. 2004, Phys. Rev. D, 70,

083511Ysard, N., Juvela, M., & Verstraete, L. 2011, A&A, 535, A89Zacchei et al. 2011, A&A, 536, A5

13


1 APC, AstroParticule et Cosmologie, Universite Paris Diderot,CNRS/IN2P3, CEA/lrfu, Observatoire de Paris, Sorbonne ParisCite, 10, rue Alice Domon et Leonie Duquet, 75205 Paris Cedex13, France

2 Aalto University Metsahovi Radio Observatory, Metsahovintie 114,FIN-02540 Kylmala, Finland

3 Agenzia Spaziale Italiana Science Data Center, c/o ESRIN, viaGalileo Galilei, Frascati, Italy

4 Agenzia Spaziale Italiana, Viale Liegi 26, Roma, Italy5 Astrophysics Group, Cavendish Laboratory, University of

Cambridge, J J Thomson Avenue, Cambridge CB3 0HE, U.K.6 CITA, University of Toronto, 60 St. George St., Toronto, ON M5S

3H8, Canada7 CNRS, IRAP, 9 Av. colonel Roche, BP 44346, F-31028 Toulouse

cedex 4, France8 California Institute of Technology, Pasadena, California, U.S.A.9 Centre of Mathematics for Applications, University of Oslo,

Blindern, Oslo, Norway10 Centro de Estudios de Fısica del Cosmos de Aragon (CEFCA),

Plaza San Juan, 1, planta 2, E-44001, Teruel, Spain11 Computational Cosmology Center, Lawrence Berkeley National

Laboratory, Berkeley, California, U.S.A.12 Consejo Superior de Investigaciones Cientıficas (CSIC), Madrid,

Spain13 DSM/Irfu/SPP, CEA-Saclay, F-91191 Gif-sur-Yvette Cedex,

France14 DTU Space, National Space Institute, Technical Universityof

Denmark, Elektrovej 327, DK-2800 Kgs. Lyngby, Denmark15 Departement de Physique Theorique, Universite de Geneve, 24,

Quai E. Ansermet,1211 Geneve 4, Switzerland16 Departamento de Fısica Fundamental, Facultad de Ciencias,

Universidad de Salamanca, 37008 Salamanca, Spain17 Departamento de Fısica, Universidad de Oviedo, Avda. Calvo

Sotelo s/n, Oviedo, Spain18 Department of Astrophysics, IMAPP, Radboud University, P.O.

Box 9010, 6500 GL Nijmegen, The Netherlands19 Department of Physics & Astronomy, University of British

Columbia, 6224 Agricultural Road, Vancouver, British Columbia,Canada

20 Department of Physics and Astronomy, Dana and David DornsifeCollege of Letter, Arts and Sciences, University of SouthernCalifornia, Los Angeles, CA 90089, U.S.A.

21 Department of Physics, Gustaf Hallstromin katu 2a, University ofHelsinki, Helsinki, Finland

22 Department of Physics, Princeton University, Princeton, NewJersey, U.S.A.

23 Department of Physics, University of California, Berkeley,California, U.S.A.

24 Department of Physics, University of California, One ShieldsAvenue, Davis, California, U.S.A.

25 Department of Physics, University of California, Santa Barbara,California, U.S.A.

26 Department of Physics, University of Illinois atUrbana-Champaign, 1110 West Green Street, Urbana, Illinois,U.S.A.

27 Department of Statistics, Purdue University, 250 N. UniversityStreet, West Lafayette, Indiana, U.S.A.

28 Dipartimento di Fisica e Astronomia G. Galilei, Universit`a degliStudi di Padova, via Marzolo 8, 35131 Padova, Italy

29 Dipartimento di Fisica, Universita La Sapienza, P. le A. Moro 2,Roma, Italy

30 Dipartimento di Fisica, Universita degli Studi di Milano,ViaCeloria, 16, Milano, Italy

31 Dipartimento di Fisica, Universita degli Studi di Trieste, via A.Valerio 2, Trieste, Italy

32 Dipartimento di Fisica, Universita di Ferrara, Via Saragat 1, 44122Ferrara, Italy

33 Dipartimento di Fisica, Universita di Roma Tor Vergata, Via dellaRicerca Scientifica, 1, Roma, Italy

34 Dipartimento di Matematica, Universita di Roma Tor Vergata, Viadella Ricerca Scientifica, 1, Roma, Italy

35 Discovery Center, Niels Bohr Institute, Blegdamsvej 17,Copenhagen, Denmark

36 Dpto. Astrofısica, Universidad de La Laguna (ULL), E-38206 LaLaguna, Tenerife, Spain

37 European Space Agency, ESAC, Planck Science Office, Caminobajo del Castillo, s/n, Urbanizacion Villafranca del Castillo,Villanueva de la Canada, Madrid, Spain

38 European Space Agency, ESTEC, Keplerlaan 1, 2201 AZNoordwijk, The Netherlands

39 Haverford College Astronomy Department, 370 Lancaster Avenue,Haverford, Pennsylvania, U.S.A.

40 Helsinki Institute of Physics, Gustaf Hallstromin katu 2, Universityof Helsinki, Helsinki, Finland

41 INAF - Osservatorio Astrofisico di Catania, Via S. Sofia 78,Catania, Italy

42 INAF - Osservatorio Astronomico di Padova, Vicolodell’Osservatorio 5, Padova, Italy

43 INAF - Osservatorio Astronomico di Roma, via di Frascati 33,Monte Porzio Catone, Italy

44 INAF - Osservatorio Astronomico di Trieste, Via G.B. Tiepolo 11,Trieste, Italy

45 INAF Istituto di Radioastronomia, Via P. Gobetti 101, 40129Bologna, Italy

46 INAF/IASF Bologna, Via Gobetti 101, Bologna, Italy47 INAF/IASF Milano, Via E. Bassini 15, Milano, Italy48 INFN, Sezione di Roma 1, Universit‘a di Roma Sapienza, Piazzale

Aldo Moro 2, 00185, Roma, Italy49 INRIA, Laboratoire de Recherche en Informatique, Universite

Paris-Sud 11, Batiment 490, 91405 Orsay Cedex, France50 IPAG: Institut de Planetologie et d’Astrophysique de Grenoble,

Universite Joseph Fourier, Grenoble 1/ CNRS-INSU, UMR 5274,Grenoble, F-38041, France

51 ISDC Data Centre for Astrophysics, University of Geneva, ch.d’Ecogia 16, Versoix, Switzerland

52 IUCAA, Post Bag 4, Ganeshkhind, Pune University Campus, Pune411 007, India

53 Imperial College London, Astrophysics group, BlackettLaboratory, Prince Consort Road, London, SW7 2AZ, U.K.

54 Infrared Processing and Analysis Center, California Institute ofTechnology, Pasadena, CA 91125, U.S.A.

55 Institut Neel, CNRS, Universite Joseph Fourier GrenobleI, 25 ruedes Martyrs, Grenoble, France

56 Institut Universitaire de France, 103, bd Saint-Michel, 75005,Paris, France

57 Institut d’Astrophysique Spatiale, CNRS (UMR8617) UniversiteParis-Sud 11, Batiment 121, Orsay, France

58 Institut d’Astrophysique de Paris, CNRS (UMR7095), 98 bisBoulevard Arago, F-75014, Paris, France

59 Institute for Space Sciences, Bucharest-Magurale, Romania60 Institute of Astronomy and Astrophysics, Academia Sinica,Taipei,

Taiwan61 Institute of Astronomy, University of Cambridge, Madingley Road,

Cambridge CB3 0HA, U.K.62 Institute of Theoretical Astrophysics, University of Oslo, Blindern,

Oslo, Norway63 Instituto de Astrofısica de Canarias, C/Vıa Lactea s/n, La Laguna,

Tenerife, Spain64 Instituto de Fısica de Cantabria (CSIC-Universidad de Cantabria),

Avda. de los Castros s/n, Santander, Spain65 Istituto di Fisica del Plasma, CNR-ENEA-EURATOM Association,

Via R. Cozzi 53, Milano, Italy66 Jet Propulsion Laboratory, California Institute of Technology, 4800

Oak Grove Drive, Pasadena, California, U.S.A.67 Jodrell Bank Centre for Astrophysics, Alan Turing Building,

School of Physics and Astronomy, The University of Manchester,Oxford Road, Manchester, M13 9PL, U.K.

68 Kavli Institute for Cosmology Cambridge, Madingley Road,Cambridge, CB3 0HA, U.K.

14


69 Kavli Institute for Theoretical Physics, University of California,Santa Barbara Kohn Hall, Santa Barbara, CA 93106, U.S.A.

70 LAL, Universite Paris-Sud, CNRS/IN2P3, Orsay, France71 LERMA, CNRS, Observatoire de Paris, 61 Avenue de

l’Observatoire, Paris, France72 Laboratoire AIM, IRFU/Service d’Astrophysique - CEA/DSM -

CNRS - Universite Paris Diderot, Bat. 709, CEA-Saclay, F-91191Gif-sur-Yvette Cedex, France

73 Laboratoire Traitement et Communication de l’Information, CNRS(UMR 5141) and Telecom ParisTech, 46 rue Barrault F-75634Paris Cedex 13, France

74 Laboratoire de Physique Subatomique et de Cosmologie,Universite Joseph Fourier Grenoble I, CNRS/IN2P3, InstitutNational Polytechnique de Grenoble, 53 rue des Martyrs, 38026Grenoble cedex, France

75 Laboratoire de Physique Theorique, Universite Paris-Sud 11& CNRS, Batiment 210, 91405 Orsay, France

76 Lawrence Berkeley National Laboratory, Berkeley, California,U.S.A.

77 Max-Planck-Institut fur Astrophysik, Karl-Schwarzschild-Str. 1,85741 Garching, Germany

78 National University of Ireland, Department of ExperimentalPhysics, Maynooth, Co. Kildare, Ireland

79 Niels Bohr Institute, Blegdamsvej 17, Copenhagen, Denmark80 Observational Cosmology, Mail Stop 367-17, California Institute

of Technology, Pasadena, CA, 91125, U.S.A.81 Optical Science Laboratory, University College London, Gower

Street, London, U.K.82 SISSA, Astrophysics Sector, via Bonomea 265, 34136, Trieste,

Italy83 School of Physics and Astronomy, Cardiff University, Queens

Buildings, The Parade, Cardiff, CF24 3AA, U.K.84 Space Sciences Laboratory, University of California, Berkeley,

California, U.S.A.85 Stanford University, Dept of Physics, Varian Physics Bldg,382 Via

Pueblo Mall, Stanford, California, U.S.A.86 UPMC Univ Paris 06, UMR7095, 98 bis Boulevard Arago,

F-75014, Paris, France87 Universitat Heidelberg, Institut fur Theoretische Astrophysik,

Albert-Uberle-Str. 2, 69120, Heidelberg, Germany88 Universite de Toulouse, UPS-OMP, IRAP, F-31028 Toulouse cedex

4, France89 University of Granada, Departamento de Fısica Teorica y del

Cosmos, Facultad de Ciencias, Granada, Spain90 University of Miami, Knight Physics Building, 1320 Campo Sano

Dr., Coral Gables, Florida, U.S.A.91 Warsaw University Observatory, Aleje Ujazdowskie 4, 00-478

Warszawa, Poland

15

intermediate results. ix. detection of the galactic haze ... · planck collaboration: detection of...

Documents